South Africa is in the midst of one of the worst epidemics of infection with the human immunodeficiency virus (HIV) recorded worldwide. The average HIV prevalence among women attending antenatal clinics nationwide in 2004 was 30%, with a population-based survey prevalence of 16.2% in 2005.1 In addition to the high prevalence of HIV, South Africa, like many developing countries, has a very high incidence of cervical cancer due to the failure to establish an effective screening program.
Infection of the cervix with high-risk types of human papillomavirus (HPV) is a prerequisite for the development of cervical cancer and its precursors. Studies have consistently shown higher prevalence of HPV infection, persistent HPV infection, and cervical intraepithelial neoplasia in HIV-positive compared with HIV-negative women.2–6 In 1995 the Centers for Disease Control and Prevention recommended that HIV-positive women should obtain two Papanicolaou (Pap) tests 6 months apart after their initial HIV diagnosis and if both results were normal, to undergo annual screening. These guidelines were, however, empirical, because little was known then about the natural history of HPV infection in HIV-positive women.
Data from an international meta-analysis of 20 HIV-positive cohorts plus the Swiss HIV cohort, show a fourfold to sixfold rise in cervical cancer incidence among HIV-positive compared with HIV-negative women.7,8 In countries where HIV infection has reached epidemic proportions, the increased risk of cervical cancer has major implications for public health.
One of the most important determinants of HPV persistence and ultimate progression to cancer is viral type, with HPV-16 being the most important. A recent meta-analysis of HPV type distribution among women infected with HIV, with data from five continents, including Africa, showed significant differences compared with women in the general population.9 These data have important implications for primary prevention of cervical cancer with HPV vaccines, which are designed to target HPV types 16 and 18.
We collaborated with investigators in Europe and the United Kingdom as part of a multicenter international cohort study,10 known as the Management of Abnormal Cytology in HIV-1 positive women (MACH-1) trial. Cross-sectional data from this study has been published.10 This article reports on the natural history of high-risk HPV infection and cervical disease in HIV-1-infected women living in Cape Town, South Africa over a 36-month period. In addition, we report on the high-risk HPV type distribution among HIV-positive women at baseline and after 18 months of enrolment, to allow an estimation of the potential effect of HPV vaccines in HIV-positive women.
MATERIALS AND METHODS
Between March 2002 and January 2003, 400 HIV-1–positive women were enrolled in a longitudinal cohort study in Cape Town, South Africa. Seventy-eight percent (n=311) were previously unscreened women recruited from a primary health care clinic, and 22% (n=89) were recruited from the colposcopy clinic at Groote Schuur Hospital, where they were referred because of an abnormal Pap test. Permission to perform the study was obtained from the Research Ethics Committee, University of Cape Town.
After providing written informed consent, all women answered a brief questionnaire and then underwent the following procedures: a conventional Pap test, sampling of the cervix for detection of high-risk HPV using Hybrid Capture II (Digene Corporation, Gaithersburg, MD), colposcopy, and histologic sampling if an acetowhite lesion was visualized. Blood was drawn for CD4 count every 6 months and HIV viral load measurements (the latter only at baseline and the 18 month follow-up visit). Pap tests were interpreted at the South African National Health Laboratory Service. For the purpose of histologic reporting, cervical intraepithelial neoplasia-1 was referred to as low-grade squamous intraepithelial lesion (LSIL) and cervical intraepithelial neoplasia-2/3 were grouped together as high-grade squamous intraepithelial lesion (HSIL).
Women with histologically confirmed HSIL were treated with large loop excision of the transformation zone. For the first two years of the study, antiretroviral therapy was not available to women in the public sector. Once antiretroviral therapy became available, eligible women (ie, women with CD4 counts of less than 200/mm3 or the diagnosis of an acquired immunodeficiency syndrome (AIDS)–defining illness) were offered antiretroviral therapy through local primary health care clinics. Thirty percent (n=121) of the cohort commenced antiretroviral therapy approximately 2 years after entry to the study.
Cervical specimens for HPV testing and typing were collected with a Digene DNA Collection Device and placed in Digene specimen transport medium (Digene Corporation). Cervical scrapings were assayed for high-risk HPV types using the Hybrid Capture II test. This assay detects 13 high-risk HPV types: 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68. The assay denaturation protocol was modified: after mixing the cervical specimen for 5 seconds, 70 microliters of the fluid was removed from the specimen tube and mixed with 35microliters of denaturation agent. The remainder of the specimen was stored at –70°C. Test results were classified as positive according to the manufacturer’s instructions. Unless otherwise stated, results reported for high-risk HPV DNA testing are as determined by the Hybrid Capture II test.
We extracted DNA using the Qiagen blood and body fluid DNA extraction kit (Qiagen Inc., Valencia, CA). Human papillomavirus genotyping was done using the Roche Linear Array HPV genotyping test (Roche Molecular Diagnostics, Pleasanton, CA). This technique detects 37 HPV types; 22 high-risk types (16, 18, 26, 31, 33, 35, 39, 45,51, 52, 53, 56, 58, 59, 66, 67, 68, 69,70, 73, 82, and IS39) and 15 low-risk types (6, 11, 40, 42, 54, 55, 61, 62, 64, 71, 72, 81, 8, 84, and CP6108). Human papillomavirus genotyping was performed on 397 women recruited at baseline and 259 women at the 18-month follow-up visit.
Cluster designation 4 enumeration was performed on a FACsCalibur flow cytometer (Becton Dickinson Inc., Franklin Lakes, NJ) using Cell Quest software (BD Biosciences, San Jose, CA). Human immunovirus-1 viral load testing was performed using the Nuclisens EasyQ HIV-1 assay (bioMérieux, Durham, NC).
Data were analyzed using Stata 9.0 (StataCorp LP, College Station, TX). All statistical tests are two tailed at α=0.05. All analyses were stratified by the source of patients (primary care clinic compared with colposcopy clinic); because no substantive differences were noted, we report only on the total cohort.
To study changes in high-risk HPV (as defined by the Hybrid Capture II test) and cervical disease through time, we divided the follow-up period into discrete intervals, with participants able to contribute data to multiple intervals. Incidence (of HPV) and progression (of cervical disease) was defined as a negative test result followed by a positive test result 6 months later. Regression (of either HPV or cervical disease) was defined as a positive test result followed by two consecutive negative test results over the following 18 months. The risk of incidence, progression, and regression of different outcomes over time are presented as percentages with exact 95% confidence intervals (CIs). We accounted for the clustering of intervals within individuals using population-average logistic regression models with the Huber-White sandwich (robust) estimator of variance.11,12 The results of these models are presented as odds ratios (ORs) with 95% CIs. In the analysis of natural history of HPV infection, study visits after women initiated antiretroviral therapy or received treatment for cervical disease were excluded. In the analysis of the effect of antiretroviral therapy on HPV status and cervical disease, we included study visits on antiretroviral therapy, but visits after cervical treatment remained excluded.
To examine the effect of baseline characteristics on the development of squamous intraepithelial lesions (SIL), we conducted survival analyses using Kaplan-Meier methods (with survival curves compared using log rank tests) and Cox proportional hazards regression modeling. For these analyses, women were censored due to death, loss to follow-up, initiation of antiretroviral therapy, and treatment of cervical disease. The results are presented as hazard ratios (HRs) with 95% CIs.
The sociodemographic characteristics and baseline HIV status of the cohort are shown in Table 1. The median CD4 count was 248/mm3. Overall, 68% (n=269) of women had a positive Hybrid Capture II test (Table 1), and the proportion of women who were Hybrid Capture II–positive did not differ in the different age groups (data not shown).
Compared with high-risk HPV–negative women, women who were high-risk HPV positive at baseline had significantly lower CD4 counts and higher viral loads (P<.001 for both associations). These were the only significant predictors of high-risk HPV positivity, with factors such as age, number of lifetime partners, current sexual activity, and smoking having no significant predictive value.
At baseline, 35% (n=137) of women had a cytologic diagnosis of LSIL, 13% (n=50) of HSIL, 7% (n=29) of ASC-US (atypical squamous cells of unknown significance), and 45% (n=176) were normal. One woman had a cytologic diagnosis of suspicious for malignancy, which was not confirmed on colposcopy or histology. Of women with ASC-US or normal cytology, 47% were high-risk HPV DNA positive, compared with 90% of women with LSIL and 94% of women with HSIL (P<.001). The proportion of women with CD4 counts less than 200/mm3 was also significantly higher in women with abnormal cytology compared with women with normal/ASC-US cytology, as were higher viral loads (Table 1).
Using a combination of colposcopic and histologic findings to evaluate cervical lesions, 59% of women had normal results (n=235), 35% (n=141) had LSIL and 6% (n=23) had HSIL. There were no cancers. Compared with women with normal colposcopy/histology findings, women diagnosed with LSIL or HSIL were significantly more likely to be high risk HPV DNA positive and to have low CD4 counts (P<.001 for each association) and higher viral loads.
At baseline, HPV typing was performed on 397 specimens, and the five most prevalent types of HPV were HPV 16 (15%), HPV 52 (15%), HPV 53 (15%), HPV 35 (14%), and HPV 18 (11%) (Table 2). Only one HPV type was identified in 27% of the specimens; 21% had two types, 12% had three types, 10% four types and 8% between five and eight types. Multiple types of HPV were found in women with lower CD4 counts and higher viral loads (P<.001, respectively). In addition, multiple types were more likely to be found among women with HSIL, determined either by cytology (Table 3) or the combined colposcopy/histology findings. For instance, 14% (n=7) of women with HSIL at Pap test at baseline had more than five types identified, compared with 12% with LSIL, 7% with ASC-US, and 1% with normal Pap tests. Human papillomavirus-16 was the commonest type identified in women with HSIL, followed by HPV-52, HPV-53, HPV-35, and HPV-18, in decreasing order of frequency (Table 3). Among women with HSIL, HPV-16 was found in 26% of cases compared with 21% with LSIL, 17% with ASC-US, and 7% with normal Pap tests.
Of the 400 women enrolled, 87 (22%) died during the 36-month follow-up period; the majority of deaths (n=76) occurred in the first 18 months when access to antiretroviral therapy was limited. By the end of the 36-month follow-up period, 30% (n=121) of the women had initiated antiretroviral therapy, most of them at or after the 24-month study visit, and 5% (n=21) had received treatment for cervical disease. An additional 97 women were lost to follow-up over the course of the study; many of these losses occurred after antiretroviral therapy initiation. Death during the study was strongly associated with lower CD4 cell counts, higher HIV viral loads, and abnormal cytology at baseline (P<.001 for all associations). Women who were lost to follow-up during the study were similar to women retained in the cohort with respect to demographics as well as HIV and cervical disease (Table 4). When person-time after antiretroviral therapy initiation and treatment for cervical disease is excluded, a total of 679 person-years of observation were included in the analysis for natural history.
During the course of the study, the 6-month incidence of high-risk HPV infection was 22% (90 new HPV infections in 403 eligible intervals; 95% CI 18–27%) among women who were initially negative for high-risk HPV. The only significant predictor of developing an incident high-risk HPV infection was low CD4 count (adjusted OR for HPV incidence with a 100-unit increase in CD4 count, 0.82; 95% CI 0.70–0.96). Clearance of high-risk HPV infection during 18 months was observed in 6% of intervals (38 of 590; 95% CI 5–9%) with a positive high-risk HPV test and two follow-up tests 6 months apart; thus, 94% of high-risk HPV infections persisted more than 18 months. Lower HIV viral load was the only significant predictor of HPV clearance (adjusted OR for a 1-log increase in viral load, 0.54; 95% CI 0.39–0.76); there was no significant association between CD4 count and clearance (adjusted OR for a 100-unit increase in CD4 count, 1.09; 95% CI 0.90–1.33).
Among women with a cytologic finding of LSIL during the study, regression to normal cytology during the following 18 months was observed in 11% of cases (41 of 388 intervals; 95% CI 8–14%). The corresponding proportion of women with cytologic diagnosis of HSIL that regressed to normal or LSIL during 18 months was 27% (14 of 51 intervals; 95% CI 16–42%). When regression of LSIL and HSIL were analyzed together as any cytologic regression, high-risk HPV status at the start of the study interval was strongly associated with regression (OR for regression among HPV-positive women compared with HPV-negative women, 0.40; 95% CI 0.18–0.89).
In those study intervals where an initial normal/ASC-US cytology result was followed by a second cytology result 6 months later, progression to LSIL or HSIL was observed in 17% of cases (108 of 647 intervals; 95% CI 14–20%). Initial findings of LSIL progressed to HSIL in 4% of cases (13 of 310 intervals; 95% CI 2–7%). When progression of LSIL and HSIL were analyzed jointly, the only significant predictor of cytologic progression was positive high-risk HPV status at the start of the study interval (OR for progression among HPV-positive women compared with HPV-negative women, 4.29; 95% CI 2.51–7.33). Similar results for both progression and regression were observed in the analysis of colposcopy/histology findings (not shown).
Among women with cytologic results of normal/ASC-US at enrollment, 42% remained so at 36 months in Kaplan-Meier analysis (95% CI 33–51%). When stratified by participant baseline characteristics, individuals who were high-risk HPV positive were much more likely to develop LSIL/HSIL compared with those who were HPV negative (Fig. 1A), as were women with low CD4 counts (Fig. 1B). In a proportional hazards model, baseline infection with high-risk HPV and low CD4 count were both independent predictors of developing LSIL/HSIL during follow-up. After adjusting for age, current sexual activity, and HIV viral load, women who where high-risk HPV positive at baseline were more than three times more likely to develop SIL than women without high-risk HPV (HR 3.09; 95% CI 1.94–4.93; P<.001). Compared with women with CD4 less than 200/mm3, women with CD4 201–500/mm3 had a 25% reduction in the hazard of SIL development (HR 0.75; 95% CI 0.45–1.26; P=.275), and women with CD4 more than 500/mm3 had a 57% reduction in the hazard of SIL development (HR 0.43; 95% CI 0.20–0.93; P=.031).
In a separate analysis of the association between high-risk HPV status and antiretroviral therapy use, there was no crude association between the use of antiretroviral therapy and high-risk HPV status (OR 1.03, 95% CI 0.77–1.4). After adjusting for age, current sexual activity, smoking, and CD4 count, there was still no association between high-risk HPV status and antiretroviral therapy. Comparing person-time on and not on antiretroviral therapy, increased CD4 count remained the most powerful predictor of being HPV negative (OR 0.50, 95% CI 0.36 – 0.71 for CD4 more than 500/mm3; OR 0.60, 95% CI0.48–0.75 for CD4 201–500/mm3). There was no significant effect of antiretroviral therapy on cytologic progression or the development of HPV infection, independent of CD4 count.
Cervical cancer remains the commonest cancer among women in developing countries, with very high incidences recorded in Southern and East Africa, where the HIV epidemic is extensive.13 This study examined the prevalence, type distribution, and risk factors for high-risk HPV infection and cervical disease in a large cohort of HIV-positive African women. As women in sub-Saharan Africa gain access to antiretroviral therapy and therefore begin to live longer, HPV-associated disease is likely to become more prevalent and clinically relevant.
The study confirms previous findings that high-risk HPV infection is highly prevalent in HIV-1–positive women, being present in 68% of participants. Women who were high-risk HPV positive at baseline were also the most immune compromised, with the lowest CD4 counts and highest viral loads. These two factors were the only significant predictors of being high-risk HPV positive, as has been demonstrated previously.14–16 In addition, this study showed a very high rate of cytologic abnormalities, with only 45% of women having normal cytology at entry. The majority had LSIL, which reflects the high rate of HPV infection, but 13% had HSIL on cytology, which is significantly higher than expected in the general population.
There seems to be considerable regional variation in HPV type distribution. In our study, the five most prevalent types were HPV 16, 52, 53, 35, and 18. In the meta-analysis by Clifford et al,9 they reported on HPV type distribution in 3,230 cytologically normal women infected with HIV. The overall prevalence of HPV in the African studies was 57% compared with 31% in Asia, 32% in Europe, 31% in North America, and 57% in South/Central America. Overall, the HPV type distribution in decreasing order of frequency was types 16, 58, 18, 52, 31, and 33. In Africa the prevalence of the different types in descending order of frequency was 16, 58, 52, 31, 18, and 35, with types 31 and 35 being significantly higher than the other continents (P<.001). These data are particularly important in understanding the potential effect of the prophylactic HPV vaccines that target types HPV 16 and 18. In considering the effect of HPV vaccination against types 16 and 18 in HIV-positive women, two points need to be made: 1) these typing data do not include any cancers, and it is possible that type distribution in cancers may still reflect types 16 and 18 as the commonest and 2) type 16 is still the commonest type, and type 18 the fifth commonest, suggesting that vaccination against these two types is likely to have a significant effect on cervical cancer prevention.
Incident HPV infection occurred in just less than one quarter of women within 6 months who were high-risk HPV negative at entry. However, once infected, the majority of women (94%) had persistent infection, and only 6% cleared the infection during the following 18 months. This is important information, because persistent infection is a known risk factor for the development of cervical cancer. Interestingly, clearance of HPV was not related to CD4 count but was to viral load. The reason for this is unclear, but we speculate that although CD4 levels in general correlate inversely with viral loads, the essential requirement for HPV clearance is functional CD8+ cytolytic activity. In this regard, CD8+ T-cell function in HIV has been shown to be impaired due to upregulation of PD-1 expression, and the latter correlates strongly with viral load and is reversed by antiretroviral therapy.17
Progression from cytologic abnormality of LSIL to HSIL only occurred in 4% of cases, whereas progression from normal/ASC-US cytology to any grade of SIL occurred in 17% of cases. Low-grade squamous intraepithelial lesions regressed to normal or ASC-US in 11% of cases. In all cases, high-risk HPV status was the most powerful predictor of cytologic progression (high-risk HPV positive) or regression (high-risk HPV negative). No cancers developed during the 3-year period of follow-up. These data suggest that in resource-restricted environments such as are found in sub-Saharan Africa, yearly screening of HIV-positive women is not necessary, and women can be safely screened at 2-year to 3-year intervals after an initial colposcopy to rule out invasive disease.
The incidence of SIL among women with normal/ASC-US cytology at baseline was strongly influenced by low CD4 count and high-risk HPV infection. It is noteworthy that only 42% of participants without SIL at enrolment remained “SIL-free” during 36 months of follow-up. Having a CD4 count of more than 500/mm3 was protective against the development of SIL for 36 months once the analysis was adjusted for age, current sexual activity, and HIV viral load. These data suggest that HIV-infected women who are immune competent will most likely behave like HIV-negative women with regard to cervical disease. Similar findings were reported by Harris et al,6 who reported that the incidence of SIL in HIV-positive women with CD4 counts more than 500/mm3 was similar to that of HIV-negative women and that similar screening practices could be applied to both groups.
This study was not designed ab initio to evaluate the effect of antiretroviral therapy on the natural history of HPV-associated disease of the cervix. However, 30% of the cohort began antiretroviral therapy within 2 years of participating in the study. We found no significant effect of antiretroviral therapy (independent of CD4 count) on the development of high-risk HPV infection or of cytologic progression, although the sample was too small to evaluate antiretroviral therapy influence on regression. This topic requires additional research because the effect of antiretroviral therapy on HPV-associated disease will have major implications for screening practices.
In summary, this 3-year follow-up study of high-risk HPV infection of the cervix and cervical cancer precursors has confirmed the strong relationship between two important sexually transmitted viruses, HIV and HPV. The expression of HPV-associated disease is strongly influenced by immune status, as reflected by CD4 counts and viral loads. Although the burden of HPV-associated disease is very high in HIV-infected women, our data suggest that biennial or triennial screening of women with normal/ASC-US or LSIL cytology would be acceptable. However, we recommend an initial colposcopy after an abnormal Pap test result to establish a baseline colposcopic and/or histologic diagnosis. Recommendations for the management of HSIL are not different from those recommended for HIV-negative women.
1. Joint United Nations Programme on AIDS/HIV. 2006 report on global AIDS epidemic. Available at: http://www.unaids.org/en/KnowledgeCentre/HIVData/GlobalReport/default.asp
. Retrieved March 26, 2008. p. 11.
2. Ferlay BF, Pisani P, Parkin DM. Globocan 2002. cancer incidence, mortality and prevalence worldwide. IARC cancer base No. 5 version 2.0. Lyon (France): IARC Press; 2004.
3. Sun XW, Kuhn L, Ellerbrock TV, Chiasson MA, Bush TJ, Wright TC Jr. Human papillomavirus infection in women infected with the human immunodeficiency virus. N Engl J Med 1997;337:1343–9.
4. Palefsky JM, Minkoff H, Kalish LA, Levine Sacks HS, Garcia P, et al. Cervicovaginal human papillomavirus infection in human immunodeficiency virus-1 (HIV)-positive and high-risk HIV-negative women. J Natl Cancer Inst 1999;91:226–36.
5. Ellerbrock TV, Chiasson MA, Bush TJ, Sun XW, Sawo D, Brudney, et al. Incidence of cervical squamous intraepithelial lesions in HIV-infected women. JAMA 2000;283:1031–7.
6. Harris TG, Burk RD, Palefsky JM, Massad LS, Bang JY, Anastos K, et al. Incidence of cervical squamous intraepithelial lesions associated with HIV serostatus, CD4 cell counts, and human papillomavirus test results. JAMA 2005;293:1471–6.
7. Clifford GM, Polesel J, Rickenbach M, Dal Maso L, Keiser O, Kofler A, et al. Cancer risk in the Swiss HIV Cohort Study: associations with immunodeficiency, smoking, and highly active antiretroviral therapy. J Natl Cancer Inst 2005;97:425–32.
8. Frisch M, Biggar RJ, Goedert JJ. Human papillomavirus-associated cancers in patients with human immunodeficiency virus infection and acquired immunodeficiency syndrome. J Natl Cancer Inst 2000;92:1500–10.
9. Clifford GM, Gonçalves MA, Franceschi S, HPV and HIV Study Group. Human papillomavirus types among women infected with HIV: a meta-analysis. AIDS 2006;20:2337–2344.
10. Kitchener H, Nelson L, Adams J, Mesher D, Sasieni P, Cubie H, et al. Colposcopy is not necessary to assess the risk to the cervix in HIV-positive women: an international cohort study of cervical pathology in HIV-1 positive women. Int J Cancer 2007;121:2484–91.
11. Huber PJ. The behavior of maximum likelihood estimates under nonstandard conditions. In: LeCam LM, Newman J, eds. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability; December 27, 1965 to January 7, 1966, Berkeley, California. Vol 1. Berkeley (CA): University of California Press; 1967. p. 221–3.
12. White H. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 1980;48:817–38.
13. Parkin DM, Bray F. Chapter 2: The burden of HPV-related cancers. Vaccine 2006;24:S11–25.
14. Parham GP, Sahasrabuddhe VV, Mwanahamuntu MH, Shepherd BE, Hicks ML, Stringer EM, et al. Prevalence and predictors of squamous intraepithelial lesions of the cervix in HIV-infected women in Lusaka, Zambia. Gynecol Oncol 2006;103:1017–22.
15. Hawes SE, Critchlow CW, Faye Niang MA, Diouf MB, Diop A, Touré P, et al. Increased risk of high-grade cervical squamous intraepithelial lesions and invasive cervical cancer among African women with human immunodeficiency virus type 1 and 2 infections. J Infect Dis 2003;188:555–63.
16. La Ruche G, You B, Mensah-Ado I, Bergeron C, Montcho C, Ramon R, et al. Human papillomavirus and human immunodeficiency virus infections: relation with cervical dysplasia-neoplasia in African women. Int J Cancer 1998;76:480–6.
17. Zhang JY, Zhang Z, Wang X, Fu JL, Yao J, Jiao Y, et al. PD-1 up-regulation is correlated with HIV-specific memory CD8+
T-cell exhaustion in typical progressors but not in long-term progressors. Blood 2007:109:4671–8.